RE: [Patch] new (almost rewritten) version of the ARM assembly ieee754 lib

["Richard Earnshaw" <Richard dot Earnshaw at arm dot com>]

> -----Original Message-----
> From: Nicolas Pitre [mailto:nico@cam.org] 
> Sent: 21 October 2004 04:20
> To: mark@codesourcery.com
> Cc: gcc-patches@gcc.gnu.org; Richard Earnshaw
> Subject: Re: [Patch] new (almost rewritten) version of the 
> ARM assembly ieee754 lib
> 
> On Wed, 13 Oct 2004, Mark Mitchell wrote:
> 
> > Nicolas Pitre wrote:
> > 
> > > On Tue, 12 Oct 2004, Mark Mitchell wrote:
> > > 
> > > 
> > > > Nicolas Pitre wrote:
> > > > 
> > > > 
> > > > > After many months of part time hacking on the ARM 
> assembly for floating
> > > > > point operations, I'm about to lose interest so it's 
> time I finally
> > > > > submit
> > > > > the improvements I've done so far and move on to 
> another distraction.
> > > > > 
> > > > > Nevertheless, the improvements in performance should 
> be significant.
> > > > > See
> > > > > for example this non-exhaustive instruction count 
> comparison between the
> > > > > old
> > > > > (current) and the new (with my patch) version of the 
> lib which should
> > > > > give a
> > > > > rough idea of the speed increase:
> > > > > 
> > > > 
> > > > Cool!  What testing have you done?  Have you run any 
> IEEE conformance
> > > > tests?
> > > 
> > > 
> > > Yes, a couple of them like paranoia, plus some of my 
> usual test tools I used
> > > when writing the original code.
> > 
> > That seems like a pretty reasonable set of tests.
> 
> Does this mean I may commit the code?
> 
> It's been more than a week without further feedback.

Sorry for the delay, it's the ARM developer's conference this week, so
I've had other things on my plate both last week and this.

Yes, this is OK

R.